General > General Technical Chat

An article on Static and Dynamic RAM.

<< < (7/7)

MK14:

--- Quote from: mawyatt on June 18, 2022, 01:51:35 am ---Intel found the source as slightly radioactive ceramic from 3M that caused the issue and instead of Intel keeping this to themselves and profiting, they let everyone know, including their competitors!!

Anyway, 3M cleaned up their ceramic and this issue was resolved.

Best,

--- End quote ---

In the old days, Intel was a pretty decent company, in many (NOT all, a good example is the 4004, where one of its original creators was ignored and left out of acknowledgement of their historic achievements, because of very badly annoying Intel, when they left it, in bad terms.  Probably  https://en.wikipedia.org/wiki/Federico_Faggin  and his creating Zilog, really upset Intel.  Although I fully accept Intel being really annoyed at the Zilog issue, and the Z80 "unfairly" stealing the limelight/profits/8080.  Intel should still have been truthful about who created/designed etc, the 4004, and onwards.).

In later (recent years), I hear less and less, about how nice/honorable Intel is as a business.  But don't get me wrong, they still seem like a decent company, over-all.  Except perhaps chasing profits too keenly, rather than doing what is best for final consumers of Intels products (such as NOT allowing/enabling ECC memory capabilities, on their cheaper CPUs, even if it is actually present on the chip.  Arguably, ECC should be enabled for EVERYONE, as it is of potential benefit to almost all users, even if your average joe probably, doesn't know what it is, or realize that they would benefit from it, due to less silent/secret/hidden data corruption).

Old DRAM chips, especially from the 1980/1990s (in my experience), but other years as well.  Seem to sometimes, just break (whereas e.g. CPUs, very rarely break, unless badly abused).  Even if just left, without being used.  I sometimes wonder why that might be.  Radiation from the package and/or background and/or cosmic-rays, possibly damaging the floating-gate (or whatever it should be called), mosfet (capacitor), and then damaging it in some way.  As, perhaps it is NOT configured to 'dissipate' the excess charge/voltage, in a safe (undamaging) way, like most other semi-conductors, typically *might*.

TL;DR
DRAMs apparently being excessively sensitive to damage/aging/destruction due to radiation from various sources.  Making them age much less well, compared to most other semiconductor ICs.

I suppose it could also be because DRAMs are really a (almost Frankenstein) mixture of digital and analogue, on the same chip.  Whereas most desktop CPUs, are mostly fully digital.  Which might compromise the processes to make those ICs, especially in huge bulk numbers and at very competitively low cost to the consumer (i.e. maybe not enough process steps, are carried out, to ensure the best chips and/or cost savings), have reduced their reliability, especially for the longer term.

Perhaps tiny changes inside the chip as it ages, are enough to 'damage/break' it, unlike CPUs, whose wide tolerance, big safety margins, and strict digital operation (not on all of them, as some of them are not truly static devices, but need minimal clocking, due to dynamic processes on some bits, to save transistors/chip area, and hence reduce costs), mean a CPU can carry on working, even if there are minor changes, in the IC, as it ages.

I've sometimes searched for answers (I suspect, but I'm NOT sure), but usually draw a blank.  Either because people don't know, or it is an industry/trade secret (the DRAM industry, seems extremely secretive, in modern times, i.e. after the DRAM industry left America), kept to the highly competitive DRAM memory suppliers.

MK14:

--- Quote from: MK14 on June 18, 2022, 05:31:58 am ---TL;DR
DRAMs apparently being excessively sensitive to damage/aging/destruction due to radiation from various sources.  Making them age much less well, compared to most other semiconductor ICs.

--- End quote ---

I shouldn't really answer my own post (question), but on reflection, I will try to.  The question seems to boil down to, why do DRAMs seem to break much more often then CPUs and other chips.

Maybe the answer is that it is because a potentially typically PCs (going back lots of years), would have usually had just one main CPU, such as an Intel or AMD one.  Whereas the main memory, would (perhaps), of consisted of 4 occupied memory slots, with each memory DRAM stick, having (maybe) 8 DRAM chips, on both sides, making a total of 16 DRAM chips per DRAM memory stick (more if parity/ECC type).

So there could have been 4 x memory DRAM sticks = 8 sides, with perhaps 8 DRAMs per side (memory capacity of DRAM chips at the time, and its pricing, dependent).
= 4 (slots) x 2 (sides per DRAM stick) x 8 (per side) = 64 DRAM chips per PC.

So, assuming (pretending) DRAM chips and CPUs have exactly the same reliability and life-expectancy per chip.  DRAM failures would be x64 times more likely to result in a failure of a PC, compared to CPUs, which there is typically only one off (ignoring all the various secondary CPUs, all over the place, potentially in PCs, such as the keyboard).

Or x32 or even x16, depending on how many DRAMs are present in the PCs main memory  e.g. only 2 slots used and/or less than 8 DRAMs per side.  Graphics typically might also have its own DRAM chips (but I'm ignoring that fact), especially if a separate part of the PC (graphics card), as opposed to iGPUs.

Navigation

[0] Message Index

[*] Previous page

There was an error while thanking
Thanking...
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod